This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
• Company presented no clinical evidence that dapagliflozin extends life, but model
predicts that dapagliflozin increased length of life, and substantially improved quality of
life → questionable face validity
• Uncertain whether model predicted what might occur after a short period of improved
glycaemic control yet an increased risk of adverse events
Consultation comments from company:
• DCCT/EDIC shows duration of diabetes and HbA1c key predictors of complications
– In company’s risk equations with other predictors e.g. blood pressure
• Risk equations for macrovascular complications from Swedish National Diabetes registry
→ robustly derived, used in other models (Sheffield and PRIME), produce results
consistent with other published risk equations:
– 1-year risk of CVD for non-smoking patient with risk factor levels of DEPICT
– is 0.4% [no history of CVD] and 1.4% [history of CVD]
vs 0.9% and 1.0% using QRisk320 and STENO19 risk equations
• Company did not have access to patient level data to derive new risk equation for
macrovascular complications from DCCT/EDIC; studies do not provide enough
information (Ontario model, Wolowacz et al, CORE)
• Scenario analyses: CVD risks from Swedish National Diabetes registry ± 20% and when
HbA1c did not affect CVD → did not change cost-effectiveness conclusions
CVD, cardiovascular disease; DCCT, Diabetes Control and Complications Trial; EDIC, Epidemiology of Diabetes Interventions and
Complications
3a. Dapagliflozin – risk of DKA
21
Committee: dapagliflozin doubles incidence of diabetic ketoacidosis and baseline
risk higher in NHS than DEPICT. Heard from clinical experts that would offer
dapagliflozin only to people who had undergone structured education
Consultation comments from company:
• DKA is serious; EMA details guidance to reduce risk
• Absolute risk low even though relative risk high – 1.7% dapagliflozin vs 1.0%
placebo
• Benefits likely to outweigh risks
• Risk in NHS similar to DEPICT → systematic review showed incidence ranges
from 8 (UK) to 51 (Germany & Austria) cases per 1,000 patient years in type 1
diabetes
• Scenario analyses increasing incidence of DKA by 2, 3, 4 or 5-fold:
dapagliflozin remains cost effective
DKA, diabetic ketoacidosis; EMA, European Medicines Agency
Which baseline risk of DKA to use in model? DEPICT or a scenario
analysis where risk is increased?
3a. Summary of product characteristics: DKA
22
“Before starting treatment with dapagliflozin:
• Assess risk factors for DKA
• Ensure ketone levels are normal. If ketones are elevated (blood beta-
hydroxybutyrate reading >0.6 mmol/L or urine ketones one plus), do not start
treatment with dapagliflozin until the ketone levels are normal
• Ensure that patient demonstrates ability to monitor ketone levels
• Patients should obtain several baseline ketone levels over 1 to 2 weeks before
starting dapagliflozin and should become familiar with how their behaviours
and circumstances affect their ketone levels
• Inform patients in a dedicated education session, on the risk of DKA, how to
recognise DKA risk factors, signs or symptoms, how and when to monitor
ketone levels and what actions to take at elevated ketone readings
• Correction of volume depletion prior to starting dapagliflozin is recommended
Committee: company underestimates costs associated with risk of DKA
DKA, diabetic ketoacidosis
3a. EMA’s advice on DKA (continued)
23
Clinical stage Blood ketones Urine ketone Actions
Ketonaemia 0.6-1.5 mmol/L Trace or
Small +
… patient should measure blood glucose
and consider taking extra carbohydrates if
glucose levels are normal or low. Ketone
levels should be measured again after 2
hours. Patient should immediately seek
medical advice and stop taking
dapagliflozin if levels persist and
symptoms present
Impending
DKA
> 1.5-3.0
mmol/L
Moderate
++
Patient should immediately seek medical
advice and stop taking dapagliflozin …
Probable DKA > 3.0 mmol/L Large to very
large
+++ / ++++
Patient should go to emergency
department without delay and stop taking
dapagliflozin
⦿ Company has tripled costs of monitoring for ketones; £238.62 applied to
dapagliflozin for increased blood glucose self-monitoring and visits – is this
sufficient?
DKA, diabetic ketoacidosis; EMA, European Medicines Agency
ERG: additional needs for education of specialists and patients not included in costs
CONFIDENTIAL
3b. Defining ‘low-insulin needs’ specified in marketing authorisation
24
Committee: “low insulin needs” not defined in marketing authorisation. Clinical
experts suggest threshold <0.5 units/kg
Consultation comments from company:
• Accepts committee’s reasons for defining ‘low insulin needs’
• Absolute DKA risk for subgroup without “low insulin needs <0.5 units/kg” is lower
than indicated population xxx vs 0.7%
• Clinically questionable → people with newly diagnosed type 1 diabetes may have
low insulin doses because of residual insulin production
– n.b. DEPICT-1 excluded people on insulin <12 months
• Difficult to implement → insulin doses vary daily
• Included in updated company base case
– small effect on cost effectiveness results
DKA, diabetic ketoacidosis
Has committee seen new evidence why it would not be appropriate
to define ‘low insulin needs’?
4. Company revised utility values – same source25
Event Used in company base case
Original Revised
Baseline utility 0.878* 0.865*
Disutilities
Cardiovascular disease 0.075** No change
Background diabetic retinopathy 0.027 No change
Proliferative diabetic retinopathy 0.070 0.029
Severe vision loss 0.074 0.059
Macular oedema 0.040 No change
Microalbuminuria 0 0.011
Macroalbuminuria 0.017 No change
Dialysis 0.169# 0.010
Renal transplant 0.023 0.010
Neuropathy 0.084 No change
Uncomplicated foot ulcer 0.125 No change
Deep foot infection 0.170 No change
Foot ulcer and critical ischaemia 0.170 No change
Minor amputation 0.117 0.080
Major amputation 0.117 0.200
Body mass index, per unit change ±0.008 0.003
Hypoglycaemia Currie equations No change
Diabetic ketoacidosis 0 0.0091
Urinary tract / genital infection 0.003 No change*Value derived from Peasgood RE model; **Weighted average of MI (-0.55; 57%), angina (-0.09; 37%) and stroke (-0.164; 7%); weightings derived
from DCCT; # Weighted average of HD (-0.164; 385 pmp) and PD (-0.204; 55 pmp); weighted by UK prevalence from Renal Registry
5. Company’s revised base case
26
Issue Company changes Committee
preference
Population insulin dose >0.5 unit/kg body weight Yes
Natural history
of disease
Progressive annual increases in HbA1c (+0.045%) and weight
(+0.1 kg) in both arms
Yes
Annual rate
stopping for
any reason
• Dapagliflozin: year 1 based on DEPICT – any reason; year
2+ based on DEPICT – adverse events
• No stopping in standard of care
Partially
Adverse
events
• Mortality: 4% DKA + 4.45% severe hypoglycaemia
• Fournier’s gangrene: £2,455 and 0.17 disutility at annual
probability 0.00001 for dapagliflozin vs 0 for standard care
Yes
Costs Dapagliflozin: ketone monitoring cost tripled + £238.62 for
increased blood glucose self-monitoring and visits
?
Utility • Disutility values from Peasgood (except amputation);
including 0.0091 for DKA
• Disutilities of 0.08 and 0.20 for minor and major amputation
Yes
Stopping rule No change – no stopping rules based on treatment
effectiveness (HbA1c) applied
No
Other Intention-to-treat dataset Yes
DKA, diabetic ketoacidosis
26
Company revised base case deterministic results, list price
27
∆ cost ∆ QALY ∆ LYICER
£/QALY
Original base case £2,026 0.31 0.32 £6,618
Revised base case £5,474 0.40 0.43 £13,775
ICER, incremental cost-effectiveness ratio; LY, life year; QALY, quality-adjusted life year
28
Scenario and sensitivity analyses
Scenarios
29
1. Risk equations
2. Waning of dapagliflozin effect
3. Stopping rules
4. Utility – different source of values
5. Company’s scenarios combined
Scenarios 1. Risk equations
30
Committee requested: analyses varying benefit of HbA1c in DEPICT including
benefit not proportional to DCCT/EDIC and no benefit after 52 week trial period
ERG comments:
• QALY gains in scenario 1 driven by weight loss
• There is a “legacy effect”* of lower HbA1c – people experience benefit in future of
having had lower HbA1c for 52 weeks
• Removing all benefits of “legacy effect” for scenario where no HbA1c or weight
benefits after 52 weeks→ ICER £119,095 (∆cost £671, ∆QALY 0.0056)
Company scenarios∆ cost ∆ QALY ∆ LY
ICER
£/QALY
1. No HbA1c effect after 52 weeks £3,791 0.20 0.24 £19,122
2. No HbA1c or weight effect after 52 weeks
(and all stop dapagliflozin)
£313 0.05 0.05 £6,385
Is a legacy effect of lower risk of complications in future from 52 weeks of
lower HbA1c plausible?
DCCT, Diabetes Control and Complications Trial; EDIC, Epidemiology of Diabetes Interventions and Complications; ICER, incremental cost-
effectiveness ratio; LY, life year; QALY, quality-adjusted life year
*At the committee meeting, the company explained that it believed that this was a factual inaccuracy
because no legacy effect has been modelled. The QALY gains in the ERG’s analysis are driven by the
differences in mortality associated with severe hypoglycaemia in year 1.
Scenarios 2. Waning of dapagliflozin’s effect
31
Committee: after initially dropping, HbA1c increases in people on (or not on)
dapagliflozin up to 52 weeks in DEPICT. Unclear if HbA1c drop with dapagliflozin
will be sustained
Consultation comments company and 1 clinical expert:
• “Commonly observed pattern in trials of effective anti-hyperglycaemic agents”:
initial drop followed by gradual increase, then long-term stabilisation
– no biological basis for waning of effect
– clinicians on advisory board suggest “most” patients who respond would
maintain benefit
– n.b. average would still increase?
– UK observational data of dapagliflozin in 5,228 patients with type 2 diabetes
show sustained treatment effect over 2 years (Prim Care Diabetes 2017;
437-444)
• Endpoints at 52 weeks were “exploratory”
Is there evidence of a waning in treatment effect?